On April 13 - Maintenance-Tips published the following Motor Testing Tip:

=========================================

"Voltage distribution on motor startup

Of electrical failures in motors, the vast majority of problems begin in the winding insulation, not the groundwall insulation. Everyone owns a meg-ohm meter but may not realize its shortcomings. A meg-ohm meter puts an even voltage across motor windings and will basically tell you if the motor is grounded or not. It is extremely limited as a predictive maintenance tool. You can drive a nail through the copper windings and not detect this with a meg-ohm meter unless it interferes with the groundwall insulation. During startup, a motor experiences voltages 3-5 times operating voltage due to contactor bouncing and other reasons for a vfd application. How many times have you witnessed a motor fail on startup? This voltage decays exponentially as it travels through the motor windings, thus causing a voltage difference between copper windings. The highest voltages are near the lead terminal and the ONLY way to duplicate this phenomenon as far as voltage amplitude and rise-time is the surge test. Surge testers output a fraction of the actual current a motor sees during startup. Find arcing problems early on with the surge test weeks or months before it turns into a shorted turn. Once a shorted turn is present, you will have a failed motor before you finish your lunch."

=========================================

We received a number of emails questioning this tip so we asked our resident independent Motor expert, Dr. Howard Penrose to clarify. We also invited other motor testing vendors to comment, however they declined.

Howard's comments:

"It starts out ‘ok,’ but becomes somewhat inaccurate. IEEE papers have proven, that repetitive high voltage testing on a new or good winding (clean and dry), has no impact on the condition of the insulation system. It is other operational and environmental issues that have an impact with the failures often occurring well within the winding more than the 2-3 turns that is evaluated with surge testing, in a large number of instances.

The claim that surge is the only way to detect these types of problems flies in the face of the independent findings of the US Coast Guard, GM, US Steel, Boeing, etc.

Yes there is some level of chatter as large motor contacts close (bounce) that causes some voltage spikes. The cause of failure on startup is actually the movement of the coils as the winding surges as shown in the video linked here. There are also quite a few independent technical papers, books, etc. that cover this issue.

Surge testing and MCA have their places and strengths and weaknesses, including the individual vendors within each technology."

=========================================

We apologize for any confusion we might have added to this important topic.

Terry O

quote:

Originally posted by Terrence O'Hanlon:
On April 13 - Maintenance-Tips published the following Motor Testing Tip:

=========================================

"Voltage distribution on motor startup

Of electrical failures in motors, the vast majority of problems begin in the winding insulation, not the groundwall insulation. Everyone owns a meg-ohm meter but may not realize its shortcomings. A meg-ohm meter puts an even voltage across motor windings and will basically tell you if the motor is grounded or not. It is extremely limited as a predictive maintenance tool. You can drive a nail through the copper windings and not detect this with a meg-ohm meter unless it interferes with the groundwall insulation. During startup, a motor experiences voltages 3-5 times operating voltage due to contactor bouncing and other reasons for a vfd application. How many times have you witnessed a motor fail on startup? This voltage decays exponentially as it travels through the motor windings, thus causing a voltage difference between copper windings. The highest voltages are near the lead terminal and the ONLY way to duplicate this phenomenon as far as voltage amplitude and rise-time is the surge test. Surge testers output a fraction of the actual current a motor sees during startup. Find arcing problems early on with the surge test weeks or months before it turns into a shorted turn. Once a shorted turn is present, you will have a failed motor before you finish your lunch."

=========================================

We received a number of emails questioning this tip so we asked our resident independent Motor expert, Dr. Howard Penrose to clarify. We also invited other motor testing vendors to comment, however they declined.

Howard's comments:

"It starts out ‘ok,’ but becomes somewhat inaccurate. IEEE papers have proven, that repetitive high voltage testing on a new or good winding (clean and dry), has no impact on the condition of the insulation system. It is other operational and environmental issues that have an impact with the failures often occurring well within the winding more than the 2-3 turns that is evaluated with surge testing, in a large number of instances.

The claim that surge is the only way to detect these types of problems flies in the face of the independent findings of the US Coast Guard, GM, US Steel, Boeing, etc.

Yes there is some level of chatter as large motor contacts close (bounce) that causes some voltage spikes. The cause of failure on startup is actually the movement of the coils as the winding surges as shown in the video linked here. There are also quite a few independent technical papers, books, etc. that cover this issue.

Surge testing and MCA have their places and strengths and weaknesses, including the individual vendors within each technology."

=========================================

We apologize for any confusion we might have added to this important topic.

Terry O

Every thing quoted in the original message about the Meg-Ohm Test is about right on. The Meg-Ohm Test can not find weak insulation turn to turn because of the reasons stated above. However; it does have predictive maintenance value. The Meg Ohm test can determine if the motor has moisture and contamination by trending the Meg-Ohm values.

The Surge Test is the only method of finding weak insulation Turn to Turn, Phase to Phase and Coil to coil because of a law of physics called Paschen's law. This is a law of physics like the law of gravity. Paschen's Law states that voltage can not bridge the gap of two bare wires with just the thickness of a hair between them until the voltage is raised above 325 volts. The surge is the only testing method that can do this between the turns.

IEEE, EPRI, Ontario Hydro and GE have all published papers on the root cause of motor stator failures. All studies indicate that the root cause analysis of 80% of motor stator failures is weak insulation turn to turn. Yes, the deterioration of this insulation is due to the mechanical movement, thermal, vibrations, torque fluctuations, and normal aging. Once the dielectric strength of the turn to turn insulation falls below the switching surges seen when an inductive circuit is broken (starting and stopping a motor) the motor will arc. This adds another failure mechanism called ozone. Eventually one of these voltage spikes will produce enough energy when coupled with the large currents associated with starting (8-10 times full load amps) and or a mechanical movement that cause the turns to fuse together. Once this happens the motor will typically fail with in minuets. To be predictive you must find the weak insulation before it becomes a turn to turn short. The only method of finding this is the Surge Test.

IEEE, EPRI, and Ontario Hydro have also studied where the failure occurs with in the windings. The study reveled that most motors will fail with in the first few turns. This is due to the high potential differences between these early turns. Yes some will fail deeper into the winding but most will be on the early turns. The surge test is not limited to the first 2-3 turns. All windings are constructed differently but the surge test has shown the ability to be above Paschen's Law as deep as half way into the winding of a phase.

A good article was written about motor failures and the recommended test to find these failures in the May issue of Uptime Magazine.

I agree we should not diminish the role of surge testing.

On the specific question of what causes a motor to fail during starting: I read above one author suggested only voltage stress and another suggested only mechanical stresses.

I would suggest that of course it can be either the mechanical stresses (movement from magnetic force) or the voltage transient. It's not necessarily the magnitude of the voltage that is the problem (although that may also be higher during start due to wave phenomena), but the steep rise-time which puts voltage stress across adjacent turns. If you have weak turn insulation, the time it's going to fail is when you stress it during a transient like starting or a voltage spike from the power supply.

Mr. O'Hanlon,

Thank you for bringing this discussion to light on the forum! At this point in time, I would like to ask the parties involved in this "predicting motor winding failure" issue to either attach actual field case studies substantiating their respective view/stand on each method of testing or give us a URL to them.

Better yet, if anyone has a documented predicted failure case study using both methods of testing technologies on the same winding, with the results of each, would be better yet!

Jim - Can you pin down which part you don't believe?

As far as test methods, megger/hi-pot tests ground insulation, surge tests primarily turn insulation (although it applies voltage to ground as well)... as mentioned before only penetrates the first few turns for form wound and further for random wound. I don't think anyone disagrees with that.

That the high starting currents produce high magnetic forces which can move the winding and cause a fault - is well known. Does anyone disagree?

The only part I added was that there is a voltage stress on turn insulation during starting. For testing, see EPRI EL 5862 Volume 1 - Executive Summary: "Using state of the art digitizers controlled by digital computers, three sets of dedicated surge monitoring equipment were developed and used on 33 utility motors (26 with air magnetic breakers and 7 with vacuum breakers). Significant surges were present only during breaker closing operations. Most surges had magnitudes from 1 to 3 p.u. and rise times from 0.2us to 0.6 us. The highest recorded surge was 4.6p.u. with 0.57 us rise time. On average, the largest magnitude surges were created by the second pole closing. Tests with and without surge capacitors confirmed that surge capacitors are effective in wavesloping waves impinging on motors..."

I believe this sufficiently supports my claim that a short-rise time surge can be present during motor starting (depends also on cable length and surge protection). I don't think anyone doubts that surges can cause turn insulation failure, do they?

This message has been edited. Last edited by: electricpete,

ElectricPete:

Actually, surges do not cause insulation failure unless the either exceed the rise time (dV/dt) rating of the insulation system or exceed the voltage withstand rating of the insulation system. Modern insulation systems have an extremely high withstand rating that is many times higher than the actual applied voltage (NEMA MG-1 and a series of IEEE papers in the IEEE Dielectrics and Electrical Insulation Society conferences). Something else has to cause weakness in the insulation system first - whether that is contamination, mechanical or other type of insulation degradation.

For instance, following a number of inaccurate papers published on the subject in the early 1990's, which stated that VFD failures in PWM drives occurred in the first few turns of the insulation system. Later research established that the actual cause of failure was partial discharge which generated ozone that ate away at the insulation system until the high electrical stresses of the PWM impulses can cross the weak spots in the insulation system.

In a little project that I performed in 2004, I used a surge comparison tester on a 480 Volt electric motor with a good insulation system and jacked the voltage up to the max of the surge tester (12,000 Volts). The motor survived a 10 second exposure with no degradation of the insulation system.

The actual question has to do with whether or not one technology or the other can detect impending failure. At this point, I am traveling, but have collected comparisons, EPRI studies, IEEE papers, and other research.

As far as the impact of degradation of insulation systems, over the past several years I was involved in a University of VA project dealing with the failure of the insulation system of the electric motor in LVAD (modern heart pumps that had a tendency of only lasting two years. The Ph.D. project was to develop concepts to increase the MTBF to 10 years.).

A copy of the paper presented at the IEEE Electrical Insulation Conference in 2005:

EIC_Initial_Insulation_Paper.pdf (93 Kb, 46 downloads)

Pete,

No, I'm not from the "Show Me State", (but perhaps should be) however I've listened to the issues and claims from the vendors claiming low voltage testing is better than high voltage testing (and visa versa) for predicting winding failures in route based predictive maintenance programs, like the issues being discussed here.

At this point I don't disbelieve or question the claims from either side, I'd just like to see the proof in the pudding, so to speak, or in this case the proof from the guy out there in the plant doing his routine route testing. Stated another way, what's working for the proactive predictive maintenance motor circuit testing technitions? Is it the low or high voltage testing or is it both? If its both, this is great! If so, in what application and what type of motor, AC/DC size and voltage?

Perhaps we can all pick up a pertinent point or two here?

Jim:

Each technology has its strengths and weaknesses. The key is to pick the one that best meets your needs.

I had the opportunity to first be trained in surge testing as an electric motor repair journeyman and field service engineer/manager. Later I ran into low voltage technologies while still in a repair shop, then again when I was at the University of Illinois at Chicago as an adjunct professor and senior research engineer. I later joined one of the manufacturers to continue research into time to failure estimation of insulation systems using low voltage testing.

Following is one article published in Maintenance Technology in 2003 (this is the full unedited GM approved edition).

General_Motors_Article.pdf (311 Kb, 42 downloads)

Following is an article that I wrote in 2000 (while I was with the vendor) based on the research at UIC that brought me into low voltage testing.

Howard

Analysis.PDF (239 Kb, 34 downloads)

And, finally, you will find an excellent article... ok, ok, I am the author... but it is pretty good... on all of the different motor diagnostic test methods, etc. that are covered by the IEEE draft standard P1415 in the July 2006 Uptime Magazine. The article covers ALL of the test methods and limits from the standard including low and high voltage testing.

And, as a shameless plug - I will be doing a 1-day workshop on all of this on September 12 at the PdM 2006 conference by request.

Howard

There are a wide variety of facts and opinions to consider for the question of surge testing vs other methods.

For the time being, I am still focused on establishing one of the many facts which are relevant to this discussion. The fact I am interested in – do motors fail from voltage surges (and do they fail from voltage surges during start).

If I understand your logic correctly Howard, you are saying that motors don’t fail from surge because 1 - motors are designed to withstand surge and 2 – you surge tested one motor at a relatively high voltage and it passed.

I don't agree with that logic. First item 1:

Yes, motors are designed to withstand surge levels well beyond those typically seen during starting.

Motors are also designed to withstand line voltage stress well beyond those seen during operation.

Motors are also designed to withstand mechanical forces from starting and effects of vibration.

Motors are also designed with various levels of ingress protection and space heating accessories to prevent failure from the adverse effects of moisture.

Even with all this design against failure, somehow motors still fail even though they appear to be properly applied and operated within their intended design envelope. There may be undetected defects from manufacturing or rewind. Over time these defects and the insulation as a whole can be degraded by factors you mentioned which may include mechanical, thermal, electrical stresses, moisture, contamination. So even if motors could withstand surges when new, they may not retain that insulation strength over time). There can also be conditions beyond design which are present that we are not aware of (moisture, contamination, abnormally severe surge environment due to faulty breaker).

Now item 2. You have successfully tested one motor and found it was resistant to failure by surge. I have tested a portion of one 13.2 kv motor winding to50 kv dc by hi-pot and 40kv surge test and didn’t see any failure. Would I be correct to conclude that all motor coils will behave this way? Not quite... not even in the same motor!. This exercise was part of a failure evaluation of a motor that had one coil failed to ground. The failed coils was isolated and we attempted to test the remainder of the coils to failure prior to rewind. The lesson is that not all coils act the same, even within the same motor (one had failed to ground at normal voltage and the others passed with flying colors at very high test voltages). Therefore I hope you’ll agree that it doesn’t sound reasonable to say that tests on one motor at one time should provide a basis for predicting failures on all motors for all times. (If only life were so easy.... we’d all be out of the testing business!)

In closing, I would like to cite two more references:
1 – Your article linked above - “An Initial Investigation of Insulation Fault effects on LVAD Motor Performance” – By Sonna M. Patel, Howard W. Penrose, Paul E. Allaire, , Zongli Lin - “Studies have shown that a leading root cause for motor failure is gradual deterioration of the insulation due to thermal, electrical, mechanical, and environmental stresses. Insulation can fault due to a voltage surge or mechanical shock due to vibration that will break down the insulation.”

2 - IEEE 522-2004 – “IEEE Guide for Testing Turn Insulation of Form-Wound Stator Coils for Alternating-Current Electric Machines” - “Experience has shown that turn insulation failures can be precipitated by abnormal steep-front surges caused by factors such as lightning strokes, faulty breaker closures, or the malfunction of various types of switching devices. However, turn insulation failures can also be caused by surges during normal breaker operations when the circuit conditions are such that the rise time of the surge at the machine terminals is less than a few microseconds.”

This message has been edited. Last edited by: electricpete,

Another reference:

“Electrical Insulation For Rotating Machines - Design, Evaluation, Aging, Testing, and Repair”

By Greg C. Stone, Edward A. Boulter, Ian Culbert, Hussein Dhirani
IEEE Press , 2004, ISBN 0-471-44506-1

“The turn insulation in form-wound coils can be exposed to very high transient voltages associated with motor starts, IFD operation, or lightning strikes. Such transient voltages may age or puncture the turn insulation.”

One more Reference:

EPRI Power Plant Reference Series 5036, Volume 16 - “Handbook to Assess the Inuslation Condition of Large Rotating Machines”
Electric Power Research Institute, 1987

“Coil turn-to-turn failures are sometimes attributed to aging of turn insulation from exposure to power system surges. The voltage between turns under normal supply conditions is relatively low, i.e. usually less than 100 Vac rms. If a fast risetime voltage surge strikes the stator winding (for example from switching on a motor), then a voltage of several kilovolts can appear across the turn insulation in the line-end coil for a short time. This voltage can puncture the turn insulation , causing a shorted turn...”

Howard,

Thanks for the nifty .pdf files. Dave Humphreys did an excellent job of assembling and presenting the GM Predictive Maintenance strategy that they have employed and it shows a lot of good work with excellent dollar payback. However, unless I missed it, I did not see one "predicted winding failure" case study with their low voltage equipment. Please correct me if I'm wrong.

I'll also have to say the same thing for the 15 page paper on "Electric Motor Energy and Reliability Analysis Using the US Department of Energy's MotorMaster+". Excellent information on energy savings, but I didn't see one example of a "predicted winding failure". Again, please correct me if I'm wrong.

Pete,

Thanks for the good replies backed up with the proper studies.

I'm a ten year veteran of 460 VAC 3 phase motor circuit condition based monitoring, using both low and high voltage motor circuit testing equipment from four different motor tester manufacturers. To date, I've predicted ONE motor winding failure with a low voltage offline resistive imbalance test. Four predicted stator mechanical/electrical problems with dynamic online testing. Several predicted rotor bar problems with online testing.

Surge testing was added the last three months to the arsenal of CBM testing and so far, two windings have failed the surge test. One has been removed from service and verified by our local motor repair shop. The other one is documented and scheduled for replacement at the next scheduled shutdown. (It's still running as it failed around 1900 volts, the other one at 1400).

Please keep in mind that I, conservatively speaking, have tested in excess of 1500 motor windings annually the last ten years with low voltage test equipment and predicted only one winding failure. On the flip side of this coin, I have confirmed many winding failures after the fact.

Jim:

First, I am going to correct something that I keep getting in emails even though I have not worked for a motor diagnostic vendor in two years! I am not the inventor of the ATPro, I joined the company in order to do research in data interpretation of a low voltage technology that I found in part of a utility funded project that outperformed the other high voltage testers and low voltage testers in head to heads that we performed. That was part of the study that we performed in one of the pdf papers that you cited.

I am supposed to be 'neutral,' in one of the groups that I am involved in, so I have to be all nice about technologies that I think, based upon my experience, are smoke and mirrors. It has called me more stress to see the BS that has resulted than I can stand any more.

I am glad that you have posted your experience with one of the technologies. I had to deal with the results of that kind of testing for years in that people did not trust low voltage testing. You are absolutely right - inductance and insulation to ground tests will not tell you very much, at all. It is unfortunate that experience then has to be extended to all low voltage testing, even though they are completely different. And, no, I will not accept you trying to compare that technology to the $1500 technology that you later purchased.

I posted the Humphrey paper so that readers could imply what was being determined. In their experience, the addition of just the low voltage testing improved their costs by 10 times when combined with all of their other technologies. He now uses it to evaluate their machine tool equipment in order to detect failing seals. The problem shows an average of six months prior to insulation short or ground failures through impedance and inductance matching. If they manage to see it as insulation to ground, it is too late. They also apply it throughout their facility including medium voltage equipment and have been able to detect impending winding faults before they have failed.

The results that they had at Allison Transmission (GM Powertrain) was then extended and a pilot project was implemented at an assembly plant. They were informed that the version of low voltage testing would find that 1 in 4 to 1 in 6 motors would have some type of electrical defect, based upon my research. The value was 1 in 4, especially in DC machines related to carbon dust incursion - not detected insulation to ground - and the problems were corrected by blowing out the DC machines. The other issues required time to determine how to prioritize the repairs. On average, we were not just detecting problems ahead of time, we were also reducing the number of mis-calls on electric motors (problems not found) reducing incorrect motor pulls by over 85% (from 15% to less than 1%). The key is, in both instances, we were detecting the issues far enough out to perform action during scheduled outages versus having to react immediately when high voltage testing is used (I used to use surge in field service and PdM applications through the 1990's).

The pilot project ended with GM corporate implementing the technology (and ESA) throughout all assembly facilities. The successes then prompted GM to continuously request (pressure the hell out of me) to consult to them. The only time I lost a sale (while I was a vendor) to a high voltage test company in a GM division was the time that my rep did not show up to the head-to-head.

Later, the US Coast Guard reviewed the technologies and after we reviewed about 10 years of data on insulation to ground testing, and were only able to find one instance where an insulation to ground test predicted a motor failure (Part of a NAVSEA RCM and MER process), they opted to evaluate low voltage testing. The results were so immediate, using the testers, that Maintenance Procedure Cards citing the particular instrument, by name, were issued and a motor diagnostics program was initiated. Hell, we were even able to detect mechanical seal failures in bow and stern thrusters using ESA! One of the original investigators has just found out that he is being transferred to Bahrain this week. He contacted me to let me know that he is taking a whole on and off line system with him as part of his agreement to take the assignment so that he can implement a program there.

In additional cases, there were enough requests that I first joined the consulting firm that took care of the Coast Guard program and also allowed me to work with GM in order to continue these programs (and so that I could also return to the overall facility and production reliability work that I have always been involved in), at their request (I have not had to make one outbound call to find consulting work!!). I finally started my own company this past January and extended my client base to US Steel for the implementation of these programs, actually using one manufacturer's low voltage tester and another's ESA equipment.

Before I left ALL-TEST Pro, I performed some research. One of the field service companies that I dealt with used to use surge testing in the field. After they changed over to low voltage testing, and performed a lot of research, I may add (I am at my office at GM in Pontiac, MI, right now and those findings are at my home office in Connecticut), comparing the two. They lent me their surge tester so that I could perform a series of experiments. I would like to remind you, as you read this, that I had already accepted my new position and was leaving the company in a few months. I have complete video tapes off all of the experiments. What was cool is that ATPro is part of BJM Pumps and I had stacks of stators that had various types of real-world failures and problems instead of the smoke and mirror setups made by most vendors for demo purposes.

I will continue this thread as soon as I return home this weekend (I am only home on the weekends) and have access to that data.

Oh, and the energy project required that we be able to detect failures before they occurred. I expected that would be inferred in both the articles.

Sincerely,
Howard

Motor_Circuit_Analysis_Concept_and_Principle_Final_Draft.pdf (293 Kb, 36 downloads)

Now, on the other subject.

I have a series of IEEE research papers, published recently, on the voltage surge issue.

Most of the surges discussed are related to surge transients in the system, not contact closing surges. These include lightning strikes, and the like, that can have a very fast rise time and voltages in excess of 10pu.

I will quote from the more recent papers from utility research during next week. I have to get my disk of papers from the IEEE Dielectrics and Electrical Insulation Society.

Howard

quote:

Originally posted by electricpete:
I agree we should not diminish the role of surge testing.

On the specific question of what causes a motor to fail during starting: I read above one author suggested only voltage stress and another suggested only mechanical stresses.

I would suggest that of course it can be either the mechanical stresses (movement from magnetic force) or the voltage transient. It's not necessarily the magnitude of the voltage that is the problem (although that may also be higher during start due to wave phenomena), but the steep rise-time which puts voltage stress across adjacent turns. If you have weak turn insulation, the time it's going to fail is when you stress it during a transient like starting or a voltage spike from the power supply.

Pete,

I concur with your analogy.

Jim

Howard,

Thanks for all the data!

Going back to my original request of case studies from the field related to "predicted winding failure" with high or low voltage test equipment, I would like to fine tune that request a bit and ask specifically for "predicted turn to turn, coil to coil or phase to phase" cases in a 460 VAC 3 phase power system, testing from the MCC such as I did in the past.